Re: scanner density-value relation

From: Richard Knoppow <dickburk@ix.netcom.com>
Subject: Re: scanner density-value relation
Date: Tue, 08 Mar 2005 21:54:21 -0800

> I wonder if flare is a factor. The test is to mask off the
> area of the step wedge so that the rest of the field is not
> illuminated.

In his case, maybe, because he reported such a drastic sensitivity to
location/direction. Mine is very different, as I tested with a step
tablet and also with large pieces of ND filters. I also compared with
the reading of a small piece of opaque stuff, forgot what, but
probably a metal foil sandwich-glued with black paper, in which some
papers and films are packaged. So I don't think it was flare or
internal reflection. If my scanner were that bad, I'd call customer
support for refund before I write here :-)

I remember writing some MATLAB codes to load the raw TIFF and do
statistical analysis, curve fitting, etc. but lost it somewhere when I
switched my computer. Now I'm interested in making a color
densitometer again to quantitatively record the effect of emulsion
formula, developer formula and toning to the image hue. Typically blue
and red reflection densities are used for those studies but I don't
have a color densitometer. With advertised Dmax of 4.0 (say that's
bogus and the linear portion goes only to 2 point something -- Warning
-- it's a number I just made up) it's still useful for routine
densitometry of pictorial negatives and prints.

By the way, if the quality of software is entirely disregarded, are
Perfection 4990 and Canoscan 9950F very closely comparable in terms of
image quality as far as b&w negs and prints are concerned? What about
usable Dmax?

(Those online reviews using dark background with fine patterns as the
test target are not very useful for this purpose because they can't
separate the noise issue and the linearity issue.)

--
Ryuji Suzuki
"Well, believing is all right, just don't let the wrong people know
what it's all about." (Bob Dylan, Need a Woman, 1982)